**About the Editors**

### **Maurizio Tiepolo**

Maurizio Tiepolo (associate professor of urban and regional planning) investigates how to improve risk management through the integration of local and scientific knowledge and public participation in the planning process. He has coordinated ten hydro-climatic risk reduction plans at the local scale. As a principal investigator, he has led several action–research and action–training projects in the Global South and has provided expertise to official development assistance.

#### **Vieri Tarchiani**

Vieri Tarchiani (researcher) investigates the impacts of climate change on the water–soil–vegetation nexus, with particular attention to semiarid environments to identify adaptation and climate risk reduction strategies and solutions. He coordinated several training and research for development projects in West Africa and collaborates with the World Meteorological Organization for the implementation and evaluation of international initiatives on climate services.

### **Alessandro Pezzoli**

Alessandro Pezzoli (senior lecturer in meteo-hydrology risk assessment and weather risk management) has a large expertise in applied meteorology and applied climatology. As an investigator on adaptation strategies to natural hazards generated by extreme weather, he participated to several research projects in Brazil, Ecuador, Ethiopia, Kenya, Niger, and Paraguay. He is a fellow of the Royal Meteorological Society and of the Royal Geographical Society.

### *Editorial* **Risk-Informed Sustainable Development in the Rural Tropics**

**Maurizio Tiepolo 1,\* , Vieri Tarchiani <sup>2</sup> and Alessandro Pezzoli <sup>1</sup>**


#### **1. Overview**

In the tropics, rural areas are still the place where many people live, despite ongoing urbanization. In tropical Africa, most of the population is still rural and will be so for at least another generation [1,2]. The development of the rural tropics is not merely a contribution to the growth of individual countries. It can be a way of reducing poverty [3,4] and inequalities in access to water [5], health care [6], and education [7] that the process of urbanization is unable to alleviate. However, it can also be a way to reduce greenhouse gas emissions that drive climate change if rural development is pursued in a sustainable way. This means stopping deforestation [8]. Then, reducing livestock-related emissions, which now account for 56%, 83%, and 87% of the greenhouse gases produced in Asia, Latin America, and Africa, respectively, according to the Food and Agricultural Organization's latest estimates [9].

Efforts to achieve sustainable rural development are often thwarted by hydro-climatic disasters (droughts, flooding, heavy rains, typhoons) which local communities are little prepared to tackle. Understanding these disasters, improving preparation, and strengthening governance have become equal priorities of the Sendai Framework for Disaster Risk Reduction [10]. However, the implementation of disaster risk reduction (DRR) at local scale, to achieve the objectives of the Sendai framework, has come across innumerable obstacles. It is often the case that agricultural practices and local planning are not very risk-informed. Climatic information is absent or not accessible locally [11]. Early warning systems and climate services are habitually not constructed with and for the rural communities [12,13]. Exposure and vulnerability are frequently considered as static determinants of risk [14]. Finally, the frequency and quality of DRR mainstreaming in local development plans are low [15,16] or simply unknown [17,18]. These deficiencies are particularly acute in the Tropics, where many Least Developed Countries are located, and where there is, however, great potential for agricultural development [19,20].

This Special Issue aims to investigate ways of increasing local knowledge of risks to contribute to rural development. It also aims to ascertain the status of essential resources, such as water and soil, and identify what undermines their integrity. Finally, it seeks to identify local policies for risk reduction and adaptation. The 22 articles collected here cover case studies from 12 countries. More than half of the articles concern Africa, as the subcontinent contains most of the Earth- s surface in the tropical zone. The 94 authors mobilized cover a wide range of disciplines, such as agronomy, architecture, civil engineering, climatology, earth sciences, ecology, economic policy, environmental engineering, geography, geology, geomatics, hydraulics, materials science, oceanography and atmospheric physics, remote sensing, and spatial and regional planning.

#### **2. A Short Review of the Contributions**

Eleven articles are devoted to the knowledge of risk. Two of them are dedicated to hazard knowledge. Vigna et al. consider the best datasets among the Climate Hazards

**Citation:** Tiepolo, M.; Tarchiani, V.; Pezzoli, A. Risk-Informed Sustainable Development in the Rural Tropics. *Sustainability* **2021**, *13*, 4179. https:// doi.org/10.3390/su13084179

Received: 30 March 2021 Accepted: 7 April 2021 Published: 9 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Group InfraRed Precipitation with Station, the Global Precipitation Climatology Center, and the Kenyan Meteorological Dataset to observe monthly rainfall trends in the North Horr subcounty in northern Kenya between 1983 and 2014. As a result, the Kenyan Meteorological Dataset corrected with a procedure based on the Global Precipitation Climatology Center monthly dataset performs better in terms of resolution and response to local scale precipitation differences.

Baldi et al. consider the severe thunderstorms in Sinai (Egypt) and their future trend. This hazard impacts the arid region as flash floods, which can be a resource if captured by water harvesting works.

Flood exposure is analyzed by Galligari et al. in a 135 km2 of the most densely populated wetlands in Niger: The Maouri temporary creek in Guéchémé. The dynamics of the built-up area in the flood zone are observed over the last ten years. Human settlements appear to be expanding by 52% in flood prone zones. House consolidation with corrugated sheet metal roofs is stronger in that zone than outside it.

Caselle et al. present a dataset to appreciate the vulnerability of local communities to drought and other threats in the Hodh Chargui region of Mauritania. The dataset is useful for drafting local development plans for the 31 municipalities that make up this vast jurisdiction.

Tiepolo et al. present a pluvial and fluvial risk assessment in four rural settlements along the Sirba River, in western Niger. The assessment is conceived to support planning risk reduction. Set up in a participatory manner, it employs innovative techniques in the region (images captured by unmanned aerial vehicles, hydraulic modelling), integrated with local knowledge. The result is a support to informed decision-making in prioritising and implementing risk reduction policies that local assessments rarely provide.

With another article, Tiepolo et al. offer a similar assessment considering meteorological, hydrological, and agricultural drought and flood risk for 13 rural communities in the Hodh Chargui, Mauritania. A large variability of risk level emerges within a relatively limited geographical area, determined by the risk of heavy rains and agricultural drought. These results are useful to identify risk reduction actions, which are very different from those usually proposed by the literature.

Frontuto et al. propose an assessment of flood impacts in Duran district, Ecuador, that considers income distribution and risk adversity instead of standard monetary damage. This influences damage compensation and the identification of priority areas for intervention.

Risk perception is the subject of the article by Gomez Zapata et al. in the case of the Cotopaxi volcano, Ecuador. The use of modelling for exposed areas identification facilitates discussion with local communities and awareness. Furthermore, it allows us to understand where exposure perceived by communities does not coincide with that calculated with the model.

Finally, three articles deal with the early warning and forecasting of extreme events. Tarchiani et al. present a locally and community-based flood early warning system designed with, and implemented for, the riverine communities along the Sirba River in western Niger. The main result is the demonstration that an early warning system can be set up operationally, involving the beneficiary communities through observation and preparedness.

Bacci et al. analyze the meteorological services delivered by the National Directorate of Meteorology to rural communities from eight municipalities of the Dosso and Tillaberi regions to reduce drought risk in agriculture. Feedback from users demonstrate the positive perception of such services and the willingness to use them, despite the intrinsic incertitude.

Ebhuoma et al. investigate the use of climate services by three rural communities in the Niger delta (Nigeria). Authors find a local preference for using indigenous knowledge rather than climate services due to the lack of local weather stations, the precedent of wrong weather predictions, and the misuse of local communication channels. A way out could be to develop pilot projects with farmers who are willing to use climate services.

Nine articles are dedicated to the state of water and soil, and the threats to these two key natural resources for rural development. Bonetto et al. consider the status of water resources in three districts of the Ethiopian Rift Valley. The study observes trends in fluoride presence, pH, and electrical conductivity values in the wells. The information obtained is useful for increasing access to drinking water in this semi-arid region.

Bertone et al. consider monitoring the presence of fertilizers on waterways in tropical Australia. The use of a mobile station for real-time monitoring of water quality, especially nitrate detection, proves advantageous over traditional laboratory sampling analysis. Nitrates have fluctuations in concentration in a short time that mobile stations can detect.

Water pollution by fecal matter is the subject of the article by Bigi et al. on North-Horr subcounty in Kenya. The presence of nitrates in water sources, and measures foreseen by local government to reduce it, highlight the greater vulnerability of the northern part of the subcounty to this threat.

Baratta et al. focus on actions to ensure greater access to water in the Kayes region of Mali. In particular, the reconstruction of damaged mini-dams with the participation of beneficiary communities is described. The restored dams increase the development of micro businesses.

Lasagna et al. consider the availability of water in Gumbo, east of Juba, South Sudan. The results of the study demonstrated the peculiarity of surface and groundwater and the critical aspects to consider for water use, particularly due to the exceeding of limits suggested by the WHO and national regulations. This research represents a first step for the improvement of water knowledge and management, for sustainable economic development, and for social progress in this African region.

Acciarri et al. consider the best option for producing drinking water at Magoodhoo island, Maldives, whose lens freshwater was damaged by the 2004 tsunami. The result is that a reverse osmosis desalination plant, powered by a photovoltaic plant with batteries, is economically and environmentally more advantageous than using a diesel desalination plant and bottled water supply.

Figueroa et al. consider the impacts of converting forests to grassland in tropical Mexico. The focus is on the dynamics of carbon, nitrogen and phosphorous balances in soil. The study observes a carbon and nutrient loss due to land use change.

Watene et al. observe soil erosion between 1990 and 2015 in the Great Rift Valley region of Kenya. Agriculture with poor soil and water conservation measures in Lake Nakuru and Bogoria–Baringo lake watersheds drive the highest erosion rates. Conservation tillage, curbing deforestation and overgrazing are recommended.

Drextler investigates climate smart agriculture adaptation in Belize. The author finds mulching, soil nutrient enrichment, and cover practices typical of Mayan farming tradition to have positive influences. However, these practices are made unsustainable by poverty, population growth and deforestation. This calls for a renewed commitment of agriculture extension services.

Two articles deal with adaptation and risk reduction at the local scale but observed in a wider context: Nigeria and tropical Africa as a whole. Ogunpaimo et al. identifies the relationship between adaptation and food security and what characteristics of rural households and farms facilitate it. The author finds that adaptation increases food security and that it is linked to access to credit and extension services.

Finally, Tiepolo et al. investigate the state of disaster risk reduction mainstreaming in local development plans for 198 rural jurisdictions over tropical Africa. Emphasis is placed on the quality of the plans rather than their number, as is done in the monitoring of the Sendai framework for DRR. Lack of climate characterization, little DRR, and low participation characterize these plans, which remain anchored in providing basic services such as as electricity, water, sanitation, and hygiene.

In the rural tropics, local communities are exposed to climate-related hazards, as well as to an unsustainable use of land and water resources. Their role in the economy and society is too important to be obscured by urban-centric policies. Support for local risk reduction should be more concerned with informing rural communities, building shared responses to face threats, and the quality of policies implemented, instead of merely considering their quantity.

The guest editors would like to thank the Italian Agency for Development Cooperation (AICS) and the ANADIA 2.0 project for their support in producing this Special Issue.

#### **References**


#### *Article*

## **Comparison and Bias-Correction of Satellite-Derived Precipitation Datasets at Local Level in Northern Kenya**

#### **Ingrid Vigna \*,**† **, Velia Bigi \*,**†**, Alessandro Pezzoli and Angelo Besana**

Interuniversity Department of Regional and Urban Studies and Planning (DIST), Politecnico di Torino & Università di Torino, 10125 Torino, Italy; alessandro.pezzoli@polito.it (A.P.); angelo.besana@unito.it (A.B.)

**\*** Correspondence: ingrid.vigna@polito.it (I.V.); velia.bigi@polito.it (V.B.)

† Both authors contributed equally to this work.

Received: 27 February 2020; Accepted: 30 March 2020; Published: 5 April 2020

**Abstract:** Understanding ongoing trends at local level is fundamental in research on climate change. However, in the Global South it is hampered by a lack of data. The scarcity of land-based observed data can be overcome through satellite-derived datasets, although performance varies according to the region. The purpose of this study is to compute the normal monthly values of precipitation for the eight main inhabited areas of North Horr Sub-County, in northern Kenya. The official decadal precipitation dataset from the Kenyan Meteorological Department (KMD), the Global Precipitation Climatology Centre (GPCC) monthly dataset and the Climate Hazards Group Infrared Precipitation with Stations (CHIRPS) monthly dataset are compared with the historical observed data by means of the most common statistical indices. The GPCC showed the best fit for the study area. The Quantile Mapping correction is applied to combine the high resolution of the KMD dataset with the high performance of the GPCC set. A new and more reliable bias-corrected monthly precipitation time series for 1983–2014 results for each location. This dataset allows a detailed description of the precipitation distribution through the year, which can be applied in the climate change adaptation and tailored territorial planning.

**Keywords:** dataset validation; precipitation; Kenya; local climate; ASALs; Quantile Mapping

#### **1. Introduction**

Over the past decades, research on climate change has become of primary concern for different disciplines at a global level. However, the understanding of the climate at a local level is key to interpreting undergoing changes. Although there is an abundance of data in the Global North, the countries of the Global South are struggling to fill the gap. More specifically, land-based meteorological stations in African countries are still around half the optimal number required, unevenly distributed and poorly equipped [1–3].

In Kenya, there are thirty-two land-based meteorological stations, distributed mainly in the south and on the coast, which are the most developed and geared towards tourism [4]. To improve the livelihoods of communities, enhance and protect property [5], the Kenyan government is promoting the country's research and development in climate information.

In North Horr Sub-County, situated in Marsabit County in northern Kenya, there are no land-based meteorological stations to provide past climate observations. At a distance of 250 km, there are three weather stations, two located in the highlands and one near lake Turkana. However, they are not close enough to describe the peculiarities of the local climate of North Horr.

The area investigated is mainly inhabited by semi-nomadic pastoral communities which rely on livestock production. They move around the area during the year looking for pasture and water according to the changing season [6]. Climate, therefore, plays an important role in their life, particularly with regard to precipitation, and extreme events such as floods, flash floods and droughts can have a catastrophic impact.

Three complex phenomena and their interaction mainly influence the climate of the Country: the Intertropical Convergence Zone (ITCZ), El Niño Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD).

Depending on the season, the periodic shift of the ITCZ north and south is mainly responsible for the bimodal rainfall pattern in Kenya. The first rainy season, known as the "long rains season", lasts approximately from March to May (MAM), and the second, the "short rains season", from October to December (OND), with some variation across the country. The ENSO and IOD can affect and alter the onset and duration of the rainy and dry seasons triggering events such as droughts and flooding [7–9]. In previous decades, changes in the amount of rainfall have been recorded. The northern Arid and Semi-Arid Lands (ASALs) region of Kenya, including Marsabit County, showed a decreasing trend. In particular, the period of 1991–2013 was generally drier than the period 1961–1990 with the MAM season having the highest, yet statistically insignificant, decline in seasonal rainfall amounts [10].

Until recently, climate reference literature for the area consisted of outdated studies [6,11,12]. However, climate change effects on climate at a local scale have increased interest in research studies of the area [13–16]. This research shows that agro-pastoralists have an awareness of climate change and that the increasing rainfall variability combined with other environmental, social and political pressures negatively affects their resilience [17–20]. However, although local knowledge is important, if it is not confirmed by official climate information, it could be unusable and ultimately useless [21,22].

The lack of land-based meteorological stations in the area requires the use of satellite-derived data and climatic models for further analysis. The relationship between large-scale weather systems and local climate varies from region to region, making necessary to evaluate and correct them at local scale [23,24], but the scarcity of land surface observation is one of the greatest difficulties in assessing dataset performances [25]. Previous studies have tried to assess the performance of satellite-derived and model-derived datasets in East Africa [26–32], in particular in Kenya [33–35], in order to address the lack of data from land-based meteorological stations. However, these studies have a more regional perspective rather than a local focus, and further investigation on their use at local scale is needed.

This study aims to contribute to precipitation data gap filling in northern Kenya through the design of an innovative methodology for the identification of the normal monthly precipitation values for the main inhabited areas of North Horr Sub-County.

Therefore, it has been necessary to assess and compare the performance of different precipitation datasets with a local scale perspective and to apply a bias correction method, leading to the creation of a new best-fit dataset for the area. Using a direct, point-to-pixel and validation through statistical indices [26,27] approach, this research compares data from models with historical data obtained from land-based meteorological stations in order to assess how well their properties fit the study area characterized by a relatively simple topography. This method was preferred to others due to its adaptability to the available data and to the study area. The no hierarchical k-means clustering method [28] was discarded because of its subjectivity. While it reduces the shortcomings caused by the differences in spatial coverage of the datasets, it requires a subjective choice on the number of clusters of pixels based on the similarity of the annual rainfall cycle. Even an analysis based on the ability of the datasets to detect rainfall events [29] was not suitable because it would have required daily precipitation data instead of monthly data.

Therefore, three model-derived precipitation datasets were selected and compared with the historical series of the nearby land-based meteorological stations of Lodwar, Marsabit and Moyale.

The precipitation datasets used were:

• The decadal dataset from the Kenyan Meteorological Department (KMD), with a resolution of 0.0375◦, hereinafter referred to as the KMD dataset [36] available at http://kmddl.meteo.go.ke: 8081/SOURCES/.KMD/;


Themost commonly used statistical indices were calculated: Bias, Mean Absolute Error, Mean Squared Deviation, Root Mean Squared Deviation, Correlation Coefficient and standard deviation [26–28,39,40]. The Taylor diagram was used as a graphical evaluation instrument [41].

The comparative analysis highlighted the relatively high performance of the GPCC dataset and the low performance of the KMD dataset. The GPCC gauge-based dataset selected was used to rectify the KMD dataset at local level on sampled reference points—the main inhabited areas, usually cited in policy planning [42,43]—using the Quantile Mapping [44] bias correction algorithm. Specific normal monthly precipitation values were identified for the reference points.

The new normal monthly precipitation values can be used in future studies for local purposes while the experimented methodology can be applied in other scant data contexts.

In Section 2, the study area is described along with the precipitation datasets that were analyzed. The steps of the methodology adopted are also detailed. In Section 3, the main results are presented. Finally, in Section 4 the conclusions are discussed with particular attention to the limits and to the possible future perspectives of the research.

#### **2. Materials and Methods**

#### *2.1. Study Area*

The study aims to define the best-fit precipitation dataset for North Horr Sub-County, which is situated in Marsabit County, northern Kenya (Figure 1). The area is considered to be part of the ASALs, with an evaporation rate that exceeds rainfall by more than ten times. However, there are some peculiarities due to the influence of the altitude on the precipitation, which makes Mt. Marsabit (1865 m above sea level), Mt. Kulal (2235 m above sea level), Hurry Hills (1685 m above sea level) and the Moyale-Sololo escarpment (up to 1400 m above sea level) quite wet areas. By contrast, the Chalbi Desert, a large salted depression lying between 435 m and 500 m above sea level, is the dryer feature of the area [45].

There are no land-based meteorological stations in the Sub-County. Therefore, an area within a 250 km radius from North Horr, the main village, has been defined and the meteorological stations located inside this area have been selected. These land-based meteorological stations are situated in Lodwar, Moyale and Marsabit.

The main inhabited areas—i.e., reference points—besides North Horr are Balesa, Dukana, El Gadhe, El-Hadi, Gus, Kalacha and Malabot.

**Figure 1.** Map of the study area: North Horr sub-County is highlighted along with the main reference points. Within the 250 km radius from the reference point of North Horr, three land-based meteorological stations are identified (Lodwar, Marsabit and Moyale).

#### *2.2. Precipitation Datasets*

Previous studies have assessed the performance of different gridded precipitation products over East Africa [32] and Kenya [34,35]. The comparison of GPCC, CHIRPS, TRMM 3B42 (Tropical Rainfall Measuring Mission) and MERRA Modern-Era Retrospective Analysis for Research and Application) based on eight major agro-ecological zones demonstrated that GPCC and CHIRPS achieved improved results in ASALs [35]. In fact, the GPCC dataset best estimates precipitation in tropical warm semiarid areas while CHIRPS best estimates precipitation in tropical warm arid areas. Similarly, the comparison of CHIRPS, TRMM 3B42, PERSIANN-CDR (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks Climate Data Record) and ARC2 (African Rainfall Climatology version 2.0) showed that CHIRPS have excellent performance in ASAL regions (high correlation, low RMSE, and low standard deviation) [34]. At regional level (East Africa), GPCC and CHIRPS have similar consistent results [32]. The other principal gridded precipitation products were evaluated. TRMM 3B42—as well as TRMM 3B43—has a reduced temporal (1998–present) [46] coverage compared to the aim of this research (1983–2013). PERSIANN-CDR underestimates rainfall in different topographical features and climatic conditions [34,47]. MERRA has a coarse resolution (0.5◦), best estimates rugged mountainous zones and inaccurately predicts the rainfall amounts in relatively low-lying areas [35]. Following these considerations, three precipitation datasets have been selected and compared (Figure 2). The reference dataset is the KMD dataset provided by the National Meteorological Service. The other two datasets where selected on the base of previous studies results. They are highly reliable because they are provided by the World Meteorological Organization and the Climate Hazard Center funded by the U.S. Agency for International Development (USAID), the National Aeronautics and Space Administration (NASA) and the National Oceanic and Atmospheric Administration (NOAA).

**Figure 2.** Comparison of the different resolutions of the gridded datasets. The Kenyan Meteorological Department (KMD) dataset has the highest resolution (0.0375◦), while the Global Precipitation Climatology Centre (GPCC) has the lowest (0.5◦). A precipitation value is available at each intersection point on the grid.

The KMD dataset is a decadal precipitation dataset, part of the Enhancing National Climate Services (ENACTS) project for development in Africa, which focuses on the creation of reliable climate data for national and local decision making. It has been produced by combining quality-controlled data from the national observation network with satellite estimates from the European Meteorological Satellites (METEOSAT). The data processing was performed using the Climate Data Tool software package developed by the International Research Institute (IRI) [36]. The dataset was directly furnished by the KMD but is also available at http://kmddl.meteo.go.ke:8081/SOURCES/.KMD/. It has a spatial resolution of 0.0375◦ and refers to the period 1983–2014.

The GPCC dataset, a monthly precipitation dataset, was developed by the Global Precipitation Climatology Project in support of the WMO World Climate Research Programme (WCRP) and the Global Energy and Water Cycle Experiment (GEWEX). It is a gauge-only product based on observations from rain gauge stations only available at a coarser resolution of 0.5◦ and a temporal coverage from 1901 to 2013 [37]. Version 7 (DOI: 10.5676/DWD\_GPCC/FD\_M\_V7\_050), available at NOAA/OAR/ESRL PSD website at https://www.esrl.noaa.gov/psd/, has been used for this study.

The CHIRPS dataset was developed to support the USAID Famine Early Warning Systems Network (FEWS NET). It builds on an high resolution and long recording period of precipitation estimates based on infrared Cold Cloud Duration (CCD) observations and on a station blending procedure based on a modified inverse distance weighting algorithm [48]. Several studies ascertain the effectiveness of this dataset in East Africa [26,27,38]. The monthly v2p0 version has been used, which is accessible through the IRI Data Library at https://iridl.ldeo.columbia.edu/SOURCES/.UCSB/.CHIRPS/. It has a spatial resolution of 0.05◦ and ranges from 1981 to near-present.

Finally, the monthly observed historical series from 1960 to 2016 from Marsabit, Moyale and Lodwar meteorological stations have been used as benchmark for the comparison analysis. They have been directly furnished by the KMD. Their characteristics are summarized in Table 1.


**Table 1.** Schematic summary of the meteorological stations' characteristics.

#### *2.3. Methodology*

The methodology followed in this study was structured in three steps (Figure 3):


**Figure 3.** Methodological scheme.

#### 2.3.1. Comparison of Dataset Performance at Meteorological Station Level

The performance of KMD, GPCC and CHIRPS datasets have been evaluated by means of a pixel to station comparison with the historical series of the meteorological stations at Marsabit, Lodwar and Moyale over a selected common period from 1983 to 2013. Five statistical indices have been computed: the bias, the Mean Absolute Error (MAE), the Mean Squared Deviation (MSD), the Root Mean Squared Deviation (RMSD) and the Correlation Coefficient (CC) [49]. A comparison of standard deviations (σ) was also performed in order to assess the dispersion of the values with regard to historical values.

A Taylor Diagram has also been created to provide an easier visual interpretation of the results. In the same graph, the CC, the RMSD, and the σ are shown for each series analyzed [41]. The MATLAB SkillMetrics toolbox developed by Peter Rochford has been used to create the diagram [50].

#### 2.3.2. Correction through the Quantile Mapping Method

The KMD dataset, besides being provided by the official National Meteorological Service, has the higher resolution and therefore it was chosen as the dataset to be corrected (D2).

The GPCC dataset was preferred to the three land-based meteorological stations as reference dataset (D1). The stations in a 250 km radius, in fact, are not representative of the ASALs, while the GPCC dataset offers reliable interpolated gauge-derived information. This approach, justified by the scarcity of observed data in the region, is in line with previous studies, which aimed to overcome this obstacle by resorting to gauge-derived datasets for the validation of satellite or reanalysis precipitation datasets [51,52]. The correction procedure aims at merging the information derived by the two datasets, namely integrating the satellite-derived data, which is found to have too low level performances, with the gauge-derived data [53,54]. Findings demonstrate that Quantile Mapping can cause inflation problems (same temporal structure and variability of the coarser grid) when applied to datasets of different resolution [55,56]. However, the procedure here used is opposite to the common downscaling procedure; in fact, it aims at correcting a high-resolution satellite-derived dataset with a coarser grid reference dataset.

The KMD dataset was therefore corrected using the Quantile Mapping bias correction algorithm technique, which has been widely used for correction of precipitation datasets [57–59] and has demonstrated high performances in arid and semi-arid areas [33,60]. In particular, the work of Ringard et al. demonstrated its usefulness for satellite-derived datasets correction in scarce observed data contexts [61]. Moreover, the Quantile Mapping correction method performs very well concerning the reproduction of the precipitation annual cycle and of the wet and dry periods length [62]. This characteristic is fundamental with reference to the monthly normal identification and for future climate analysis of the area.

The Quantile Mapping technique is based on statistical transformation which attempts to adjust the distribution of modelled data such that it closely resembles the observed climatology solved using a theoretical distribution The 'qmap' package developed by Lukas Gudmundsson for R software was used for the computation [63]. The procedure was carried out for all the reference locations using a pixel to pixel approach. The 'qmap' package supports five different analytical methods, both parametric and non-parametric transformations. These methods use different functions to transform the distribution of the modelled data to match the distribution of the observations. The five functions performed are parametric transformations (PTF), distribution derived transformations (DIST), non-parametric quantile mapping using empirical quantiles (QUANT), non-parametric quantile mapping using robust empirical quantiles (RQUANT) and quantile mapping using a smoothing spline (SSPLIN)(for further details see the documentation at the following link https://www.rdocumentation.org/packages/CSTools/ versions/2.0.0/topics/CST\_QuantileMapping).

Five precipitation series were created for the three stations and for the eight reference locations.

The results of the Quantile Mapping correction were compared with the GPCC series for each reference location through the statistical indices (Section 2.3.1). The most appropriate method was identified, leading to the selection of a best-fit precipitation series for each reference location.

#### 2.3.3. Reference Values Computation

New reference values were computed on the new precipitation series by averaging the monthly precipitation amount for the entire period (1983–2013).

#### **3. Results and Discussion**

#### *3.1. Comparison of Dataset Performance at Meteorological Station Level*

The comparison of the precipitation datasets with the observed series led to an important first conclusion. The KMD dataset does not feature the best indices values for all the stations. Results from the first step of the analysis conducted indicated that the GPCC dataset was a better choice as the reference series.

According to the statistical indices (Tables 2 and 3) and to the Taylor diagrams (Figure 4), the GPCC dataset fits better for the stations of Marsabit and Moyale, while the KMD dataset fits better for Lodwar. However, for reasons of homogeneity and consistency, the GPCC dataset was also chosen as the reference dataset for Lodwar station since its statistical values are close to the values obtained for the KMD dataset.



MSD **237** 260 414 1498 **933** 1528 2797 **966** 1087 RMSD **15.4** 16.1 20.35 38.71 **30.5** 39.09 52.9 **31.1** 32.97 CC 0.83 \*\*\* **0.85 \*\*\*** 0.71 \*\*\* 0.91 \*\*\* **0.95 \*\*\*** 0.92 \*\*\* 0.82 \*\*\* **0.91 \*\*\*** 0.90 \*\*\*

**Table 3.** Comparison based on the standard deviation of the precipitation datasets with the observed historical series from the selected land-based meteorological stations (Lodwar, Marsabit and Moyale) for the period 1983–2013.


**Figure 4.** Taylor diagrams showing the agreement between the observed historical series and the precipitation datasets for the selected land-based meteorological stations (Lodwar, Marsabit and Moyale) for the period 1983–2013. The standard deviation of each series (as reported in Table 3) is proportional to the distance from the origin of the diagram. The Correlation Coefficient (in Table 2) between each series and the observed historical series is expressed by the azimuthal angle. Finally, the Root Mean Squared Deviation (in Table 2) between each series and the observed historical series is proportional to the distance from the point representing the observed historical series. Points closer to the historical series' marker, that is, with similar standard deviation, lower RMSD and higher Correlation Coefficient, correspond to the best-fit datasets.

#### *3.2. Correction through the Quantile Mapping Method*

As showed in the previous section, the GPCC dataset fits better than the other two datasets compared with the historical series. However, the GPCC dataset has a lower resolution (0.5◦) compared to the KMD dataset and CHIRPS dataset (0.0375◦ and 0.05◦, respectively), and the differences in local topography may be biased. Therefore, it was necessary to apply a bias correction method to overcome these two problems. The strategy adopted was to correct the KMD dataset (D2), which is issued by the official National Meteorological Service and has the highest resolution, with the GPCC dataset (D1) which performs better on ASALs.

The bias correction method is performed using the Quantile Mapping method from the 'qmap' R package. Five different bias-corrected series are obtained based on the five different transformations applied: Parametric Transformations (PTF), Distribution Derived Transformations (DIST), Robust Empirical Quantiles (RQUANT), Empirical Quantiles (QUANT), Smoothing Spline (SSPLIN).

From comparison analysis, the Parametric Transformations method, which fits a parametric transformation to the quantile-quantile relation of observed and modelled values, provided the best results (see Appendices A and B). Hereinafter, the Bias-Corrected KMD dataset will be referred to as the BCKMD dataset (D3).

#### 3.2.1. Quantile Mapping Validation at Station Level

The performance of the new BCKMD dataset is assessed by means of the statistical indices mentioned previously (see Section 3.1). The indices have been calculated in relation to the historical series of the land-based meteorological stations, then compared with the same indices calculated for the KMD dataset.

As shown in Table 4, the BCKMD dataset fits the observed historical series better than the KMD dataset, apart from Lodwar station. This may be due to a higher performance of the KMD dataset—before correction—at Lodwar station compared to the GPCC concerning BIAS, MAE, MSD and RMSD. However, the errors obtained are still acceptably low. In fact, the standard deviation values and the relatively low values of the error's indices, even for Lodwar, justify the selection of the BCKMD dataset for the study area.

**Table 4.** Comparison based on statistical indices (BIAS, MAE, MSD, RMSD, CC and σ) of the KMD and of the Bias-Corrected KMD (BCKMD) datasets with the observed historical series from the selected land-based meteorological stations (Lodwar, Marsabit and Moyale) for the period 1983–2013. For the CC index, "\*" corresponds to a *p*-value < 0.01, "\*\*" corresponds to a *p*-value < 0.001 and "\*\*\*" corresponds to a *p*-value < 0.0001. Values in bold correspond to the best value of the index for each station.


#### 3.2.2. Quantile Mapping at Reference Point Level

The Quantile Mapping correction on the base of the GPCC dataset was also applied to the KMD dataset at the reference points. The Parametric Transformations method has been used in accordance with the validation carried out at the stations level. The result was a best-fit precipitation dataset for the eight locations.

#### *3.3. Calculating Normal Values of Precipitation at Station Level*

The normal values were calculated on the precipitation series obtained for the three stations, by averaging the monthly precipitation amount for the entire period (1983–2013) (reported in Table 5). Long rains amount, short rains amount and total annual amount were also calculated. Figure 5 compares the distribution of the precipitation through the year according to the observed series and to the new BCKMD dataset.

**Table 5.** Comparison of normal values of precipitation (in mm) obtained from the historical time series and the new precipitation dataset BCKMD for the three land-based meteorological stations (Lodwar, Marsabit and Moyale). The table reports monthly cumulative amounts, long rains cumulative amount, short rains cumulative amount and total precipitation amount.


**Figure 5.** (**a**) Representation at local scale of the historical time series and of new bias-corrected monthly precipitation time series for each land-based meteorological station divided into three cumulative amounts: total annual amount (black and very dark blue bars), long rain season precipitation amount (dark grey and dark blue bars) and short rain season precipitation amount (light grey and light blue bars). (**b**) Comparison of the annual precipitation distribution for each station according to the observed series and to the new bias-corrected monthly precipitation dataset.

#### *3.4. Calculating Normal Values of Precipitation at Reference Point Level*

The normal values were calculated on the precipitation series obtained for each reference point, by averaging the monthly precipitation amount for the entire period (1983–2013). Moreover, long rains amount, short rains amount and total amount were calculated.

The normal values for the eight reference points are shown in Table 6. A visual representation of the precipitation distribution at local scale is pictured in Figure 6.

The understanding of climate differences at local scale is crucial for an effective territorial planning against negative impact of climate change. This study succeeded in obtaining normal values of precipitation for each reference point despite the lack of land-based meteorological stations in the area and high-resolution and fitting satellite-derived precipitation time series. Differences in rainfall regime are evident in Figure 6, which shows higher precipitation amounts in the northern part of the Sub-County then in the southern reference points.

The new precipitation time series can be used for the evaluation of drought indices as well as for water security assessment. More specifically, the monthly normal values can be used as reference values for comparing measured or forecasted data in order to evaluate drought or wet periods.

Moreover, it has been possible to calculate the normal values for the entire long rain season and short rain season, by cumulating monthly values for March, April and May and for October, November and December, respectively. Knowing the distribution of the precipitation throughout the year and the possible deviation from normal values is fundamental. This is at the base of the community organization for the local semi-nomadic pastoral population.

**Figure 6.** (**a**) Representation at local scale of the new bias-corrected monthly precipitation time series for each reference point divided into three cumulative amounts: total annual amount (black bar), long rain season precipitation amount (dark grey bar) and short rain season precipitation amount (light grey bar). (**b**) Annual precipitation distribution for each reference points based on the new bias-corrected monthly precipitation time series.


**Table 6.** Normal values of precipitation (in mm) for the eight reference points in North Horr Sub-County. The table reports monthly cumulative amounts, long rains cumulative amount, short rains cumulative amount and total precipitation amount for each reference point.

#### **4. Conclusions**

The aim of this study was to obtain the normal values of the monthly amount of precipitation for the main inhabited areas in North Horr Sub-County, in order to provide a benchmark for understanding the ongoing changes in the local climate. Therefore, it was necessary to identify an appropriate historical precipitation series. The comparison between the GPCC, KMD and CHIRPS datasets highlighted the lower performance of the KMD dataset compared to the others, despite it being the dataset officially issued and used by the Kenyan Meteorological Department for the whole country. Previous studies on East Africa indicated the CHIRPS dataset to be a reliable global dataset for the region [26,27,32]. The relatively high performance of the GPCC dataset in northern arid Kenya is in line with the results of previous studies, which indicated it as a good fit for the ASALs [35], but with a low capacity in representing complex terrain [28]. However, the need to highlight local differences in the annual trend of precipitation led to the use of the KMD dataset after a correction procedure based on the GPCC dataset. This approach aimed to integrate the higher resolution of the KMD dataset—namely, its ability to detect differences in the precipitation trend at a local scale—with the higher ability of the GPCC dataset to represent the real historical values in the area. The methodology adopted created a new bias-corrected monthly precipitation time series for each reference point, from which the local normal values were extracted.

Since the need for high-resolution precipitation data covering the Global South is becoming urgent for any discipline that must consider the role of climate, this study represents an attempt to provide a solution to the scarcity of observed data. The absence of land-based meteorological stations in the area, however, cannot be ignored and constitutes a limit in the study. Future research should be directed to test the methodology proposed here in other contexts, where the availability of observed data could provide a yardstick for its usefulness and accuracy.

**Author Contributions:** Conceptualization, A.P., V.B. and I.V.; methodology, V.B. and I.V.; formal analysis, V.B. and I.V.; writing—original draft preparation, V.B. and I.V.; writing—review and editing A.P., V.B., I.V. and A.B.; supervision, A.P. and A.B. All authors have read and agreed to the published version of the manuscript.

**Acknowledgments:** This study was conducted within the framework of the International Cooperation Project "ONE HEALTH: Multidisciplinary approach to promote the health and resilience of shepherds" communities in North Kenya" funded by the Italian Agency for Development Cooperation (AICS). The authors would like to thank the project coordinator (CCM) and project partners (TRIM and VSF-Germany) and the Kenyan Meteorological Department.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

**Table A1.** Comparison based on statistical indices (BIAS, MAE, MSD, RMSD, CC and σ) of the GPCC dataset with the five bias-corrected series for the period 1983–2013. For the CC index, "**\***" corresponds to a *p*-value < 0.01, "**\*\***" corresponds to a *p*-value < 0.001 and "**\*\*\***" corresponds to a *p*-value < 0.0001. Values in bold correspond to the best value of the index for each reference point.



**Table A1.** *Cont.*

#### **Appendix B**

**Table A2.** Comparison based on statistical indices (BIAS, MAE, MSD, RMSD, CC and σ) of the historical values with the five bias-corrected series (PTf, DIST, RQUANT, QUANT, SSPLIN) for the period 1983–2013. For the CC index, "**\***" corresponds to a *p*-value < 0.01, "**\*\***" corresponds to a *p*-value < 0.001 and "**\*\*\***" corresponds to a *p*-value < 0.0001. Values in bold correspond to the best value of the index for each station.



**Table A2.** *Cont.*

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Climatology and Dynamical Evolution of Extreme Rainfall Events in the Sinai Peninsula—Egypt**

#### **Marina Baldi 1,\* , Doaa Amin 2, Islam Sabry Al Zayed <sup>3</sup> and Giovannangelo Dalu 1,4**


Received: 7 June 2020; Accepted: 28 July 2020; Published: 31 July 2020

**Abstract:** The whole Mediterranean is suffering today because of climate changes, with projections of more severe impacts predicted for the coming decades. Egypt, on the southeastern flank of the Mediterranean Sea, is facing many challenges for water and food security, further exacerbated by the arid climate conditions. The Nile River represents the largest freshwater resource for the country, with a minor contribution coming from rainfall and from non-renewable groundwater aquifers. In more recent years, another important source is represented by non-conventional sources, such as treated wastewater reuse and desalination; these water resources are increasingly becoming valuable additional contributors to water availability. Moreover, although rainfall is scarce in Egypt, studies have shown that rainfall and flash floods can become an additional available source of water in the future. While presently rare, heavy rainfalls and flash floods are responsible for huge losses of lives and infrastructure especially in parts of the country, such as in the Sinai Peninsula. Despite the harsh climate, water from these events, when opportunely conveyed and treated, can represent a precious source of freshwater for small communities of Bedouins. In this work, rainfall climatology and flash flood events are presented, together with a discussion about the dynamics of some selected episodes and indications about future climate scenarios. Results can be used to evaluate the water harvesting potential in a region where water is scarce, also providing indications for improving the weather forecast. Basic information needed for identifying possible risks for population and infrastructures, when fed into hydrological models, could help to evaluate the flash flood water volumes at the outlets of the effective watershed(s). This valuable information will help policymakers and local governments to define strategies and measures for water harvesting and/or protection works.

**Keywords:** Sinai Peninsula; flash flood; climate change; CORDEX; water harvesting

#### **1. Introduction**

In recent decades, as in past times, heavy rainfall resulting in flash floods has affected not only the Egyptian coastal areas along the Mediterranean Sea and the Red Sea but also arid and semi-arid areas such as Upper Egypt (e.g., Luxor, Aswan, and Assiut) and the Sinai Peninsula [1–5].

Under changing climate conditions, the frequency of extreme rainfall episodes is also changing. These episodes, although rare, can be catastrophic in regions characterized by very low annual precipitation, with large impacts on lives, infrastructures, properties, and last but not least, to the great cultural heritage of Egypt [6]. Recent heavy rainfall events in the densely populated region of Cairo Governorate have forced authorities to close schools, offices, and highways connecting Cairo to

other provinces, producing electricity disruption and floods in large areas of the town. These floods have also trapped people in their cars for several hours, as happened during the recent episodes that occurred in April 2018, October 2019, and March 2020 [7–10].

Scientific literature [11] has illustrated how the heavy rainfall and subsequent flash floods can be harmful in the Sinai Peninsula and how important the production of risk maps and implementing the flash flood protection works for the main watersheds of the Peninsula is, taking into consideration the complex morphological parameters [12,13]. Concerning the heavy rainfall episodes that have occurred in the Middle East Region and specifically in Egypt, many authors, e.g., [14–18], have focused their analysis on the evolution of extreme rainfall episodes and on the atmospheric pattern precursors of the evolution of these events. Due to the scale of the rare phenomenon, the forecast of a small or, at its largest, mesoscale severe thunderstorm, bringing heavy rainfall and causing flash flood, is not an easy task. Only the use of an appropriate tool like a mesoscale model can help to produce a timely and detailed forecast of the episode, which can be used as the major "ingredient", together with data-driven weather-runoff forecast models, for an Early Warning System (EWS) to be adopted in order to avoid or, at least, minimize major impacts and save lives [19–21]. In addition, if, in this flood-prone region, it is important on one hand to minimize damages and avoid disastrous events deriving from severe thunderstorms, on the other hand, it is also important to favor the use of rainfall as a precious source of fresh water. For this purpose, it is essential that the development and use of EWS [19–21] present high performances.

In order to decrease the uncertainties in the EWS, it is necessary to work on the improvement of Weather Prediction Tools (WPT) in parallel with the adoption of a sophisticated data-driven weather-runoff model; however, this improvement can come only through a thorough validation of the WPT. Due to their randomness, low frequency of occurrence, rapidity in evolution, and small mesoscale dimension, observations related to these thunderstorms are scarce in the Region and do not permit a robust validation of WPT and therefore of EWS. In this respect, the aim of the present study is to contribute to an increase of the knowledge about the present and future occurrence of extreme rainfall events in the Sinai Peninsula and on their dynamics.

To this aim, the study starts with a general overview of the climatic trends in the Sinai Peninsula, followed by a discussion on the climatology of extreme rainfall events in Sinai and by the analysis of a selection of heavy rainfall episodes. The second part of the study presents a general overview of future climate scenarios that might affect the frequency and intensity of extreme events in this region. This part of the work not only gives some general information, but also results can represent a necessary basis for more detailed climate scenario studies, and it will possibly encourage the development of specific regional projections at a higher resolution. More detailed scenario analysis can be used for the implementation of more sophisticated risk maps for future decades and can foster the elaboration and adoption of adaptation plans for population, agriculture, and water resources management and, finally, might help the decision-makers to plan the construction of flood water harvesting structures and flash flood protection works.

#### **2. Materials and Methods**

#### *2.1. Data and Methods*

The climatological analysis of the current climate and its changes in the last decades is based on the National Centers for Environmental Modeling/National Center for Atmospheric Research global analyses of atmospheric fields, NCEP/NCAR reanalysis, provided by National Oceanic and Atmospheric Administration (NOAA) [22] as well as the global atmospheric reanalysis ERA-Interim provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) available at a spatial resolution of approximately 80 km [23]. The annual rainfall trend over Sinai Peninsula is analyzed using the Global Precipitation Climate Center data (GPCC) provided by the NOAA Physical Sciences Laboratory (PSL), Boulder, Colorado, USA (https://psl.noaa.gov/). The study of the spatio-temporal evolution of the selected episodes of heavy rainfall is based on the analysis of meteorological data from ground stations in Sinai and rainfall estimates derived by satellite images from Tropical Rainfall Measuring Mission (TRMM) and the Hydrologic Data and Information System (HyDIS). In addition, NCEP-NCAR reanalysis is used to highlight the atmospheric patterns before, during, and after the selected episodes in order to individuate (i) the main atmospheric patterns contributing to the evolution of the events, (ii) possible driving mechanisms, and (iii) similarities among the events. Finally, climate scenarios at the regional scale were obtained from the outputs available through the Coordinated Regional Downscaling Experiment (CORDEX).

#### *2.2. Study Area*

Egypt is located on the northeastern side of Africa, covering nearly 3% of the total area of the continent. This country is bordered to the north by the Mediterranean Sea and to the east by the Red Sea, by the Gulf of Aqaba and Palestine, to the south by the Republic of Sudan, and to the west by the Republic of Libya. A few geographical sub-regions of the country can be identified: the Nile valley and delta, covering about 4% of the total area; the Eastern desert, covering about 22%; the Western desert, covering about 68%; and the Sinai Peninsula, covering about 6%.

Hot and dry conditions characterize the general climate of Egypt with an average annual rainfall over the whole country of about 10 mm. Even along the narrow northern strip of the Mediterranean coastal land, where most of the rainfall occurs, the average annual rainfall is usually less than 200 mm, and this amount very rapidly decreases proceeding further inland. The only regular supply of water is from the Nile. This river, which has its main source in the highlands of East Africa, crosses Egypt from the south to north end with a large delta in the Mediterranean Sea. In Egypt, the Nile water is artificially channeled on both sides. The Nile supports the life along its length, and allowed the growth of a great civilization in peace and stability [24]. The importance of the river was very clear since ancient times; in fact, Herodotus (484–425 BC) stated that "Egypt is the Gift of the Nile".

The study area of the present work is Sinai, a triangular-shaped peninsula with an area of about 60,000 km2; this peninsula is bounded by the Mediterranean Sea to the north, the Red Sea to the south, and by the Gulfs of Suez and Aqaba to the west and to the east, respectively (Figure 1). The territory of the region is quite rough with a complex orography and elevated mountains, which reach up to and above 2400 m ASL (above sea level); Mount Catherine, Egypt's highest mountain, reaches 2642 m ASL. The central area of Sinai consists of two plateaus, Al-Tih and Al-Ajmah, both deeply indented and dipping northward towards Wadi El Arish. Towards the Mediterranean Sea, the region is characterized by a plateau, by a system of a number of dome-shaped hills, and by a belt of parallel dunes, which can be as high as 100 m.

**Figure 1.** The location of the study area showing the Sinai Peninsula and its orography. Credits: ESRI World Imagery.

The whole territory of the Peninsula is characterized by the presence of numerous watersheds, which discharge rainwater on three sides: the Mediterranean Sea, the Gulf of Aqaba, and the Gulf of Suez (Figure 2). Rivers in these watersheds are short but fierce because of the combined effect of steep slopes and rare episodes of heavy rainfall. The main characteristics of the watersheds in the Sinai Peninsula are summarized in Table 1, as in [25].

**Figure 2.** Watersheds of study area (Source: Elsayed et al., 2013).


**Table 1.** The geomorphology characteristics of the watersheds of Sinai (Source: [25]).

Sinai is economically important for Egypt because it represents one of its largest mining areas. This peninsula has also a great potential for agriculture and industrial development. Although there are some limitation: the region is prone to flash flood events resulting from heavy, sudden, and short duration rainfall events, which represent a risk for the population, infrastructures, properties, and economic sectors like industry and agriculture itself.

On the other hand, flash floods caused by heavy rainfall events in Sinai and southern/southeastern Egypt represent a potential source of non-conventional freshwater resources. In particular, the water, which usually drains into the Gulf of Suez and the Gulf of Aqaba, if wisely harvested, could fulfill a non-negligible amount of water demand and/or recharge the shallow groundwater aquifers, becoming a precious fresh water source for local people of the region and their agriculture.

#### **3. Results**

#### *3.1. Climate Variability in the Sinai Peninsula*

The temperature and precipitation monthly mean over the country are shown in Figure 3, evaluated using ECMWF ERA-Interim reanalysis for the period 1981–2010. In the northern part, the Sinai Peninsula is characterized by a Mediterranean climate, whilst in the southern part, the climate is semi-desert to desert. Thus, in most of the Peninsula the climate is hot to very hot. However, sub-regions along the Mediterranean coast in the North and over the mountains are more temperate. In winter, the temperatures are surely lower, with temperatures that can drop to 0 ◦C at night over the mountains (Figure 3). Most of the precipitation occurs during the winter and then in spring and autumn, while during summer rainfall is almost totally absent over the Sinai (Figure 3).

Using ECMWF ERA-Interim reanalysis, an increase of about 1 ◦C in mean temperature has been evaluated over Sinai, as shown in Figure 4. This figure shows the anomalies of air temperature at 850 hPa during the last decade over the African Continent at large compared to the baseline period (1979–2000). In addition, the analysis performed using ECMWF ERA-Interim reanalysis for the period 1979 to 2018 over the domain (27◦–32◦ N, 32◦–35◦ E) shows a clear tendency towards increasing average temperatures (Figure 5) and decreasing total rainfall (Figure 6).

#### *Sustainability* **2020**, *12*, 6186

**Figure 3.** Climatology of Egypt, including the Sinai Peninsula, evaluated over the period 1981–2010. Monthly mean precipitation (in mm) and air temperature (in ◦K) at2m.(DataSource:EuropeanCentreforMedium-RangeWeatherForecasts(ECMWF)ERA-Interimreanalysis).

**Figure 4.** Mean air Temperature at 850 hPa: anomaly over the period 2010–2018 relative to the base period 1979–2000 (Data Source: ECMWF ERA-Interim reanalysis, available at: https://climatereanalyzer.org).

**Figure 5.** Annual and seasonal mean (top panel) and anomalies (bottom panel) of air temperature at 2 m over the Sinai Peninsula for the period 1979–2018, evaluated in the domain (27◦–32◦ N, 32◦–35◦ E). Base period for anomalies (1979–2000). (Data Source: ECMWF ERA-Interim reanalysis, available at: https://ClimateReanalyzer.org).

**Figure 6.** Annual Total Precipitation over the Sinai Peninsula for the period 1979–2018, evaluated in the domain (27◦–32◦ N, 32◦–35◦ E). (Data Source: ECMWF ERA-Interim reanalysis, available at: https://ClimateReanalyzer.org).

The analysis of the GPCC historical data for the period 1901–2013, shows a tendency to decrease precipitation in different sub-regions of the Sinai Peninsula: in the North, at the border with the Mediterranean Sea, in the Centre, and in the South (Figure 7). In general, the total rainfall distribution is quite different along the Peninsula, and it decreases from the northeast towards the southwest, with a maximum in Rafah. The southern part of the Peninsula is characterized by much lower rainfall totals in coastal areas, along the Gulfs of Suez and Aqaba.

**Figure 7.** Annual Total Precipitation in North, Middle, and South Sinai Peninsula with linear decreasing trend superimposed. (Data Source: Global Precipitation Climate Center data (GPCC) —https://www.esrl.noaa.gov/psd/data/gridded/data.gpcc.html).

#### *3.2. Extreme Rainfall Events in Sinai*

The analysis of rainfall data is the most important element in hydrological studies because it can be used in the EWS development as input to determine the flood discharges. Timing, duration, and extent of rainfall episodes are especially important in Sinai because they represent the major potential source of renewable water. Unfortunately, these flash floods are very difficult to forecast; in

addition, their fresh water is difficult to collect because of the steepness of the slopes of the terrain in the peninsula. Moreover, in this semiarid-to-arid region, rainfall is usually characterized by extremely high spatial and temporal variability [26], with intense episodes of very short duration, spatially scattered over the territory. These characteristics make the collection of this water very difficult without the construction of important infrastructures. An additional source of fresh water is the non-renewable groundwater, as in the case of the Nubian Aquifer.

As already mentioned, in recent decades Sinai has been affected by significant changes in the climate, with drier and hottest conditions relative to the baseline period (Figures 4–6), and with more intense storms with associated short, but intense rainfall events. This trend caused severe and long dry periods suddenly interrupted by sporadic intense rainfall with increased space-time variability; this has increased the forecast difficulties in recent years. If this tendency continues, the impacts on the natural environment and resources, including renewable water, as well as on population, infrastructure, and properties, will be severe.

The major storms which affected North and South Sinai in the past years are listed in Table 2 for South Sinai and in Table 3 for North Sinai. The lists show how the number of events since the year 2000 is quite significant.


**Table 2.** List of heavy rainfall events in the South Sinai Peninsula at different locations.

<sup>1</sup> Other events occurred in South Sinai more recently, including a 27–29 October 2016 episode when the storm also affected the Red Sea, Ismailia, Beni Suef, Qena, Assiut, and Suhag provinces (https://reliefweb.int/disaster/fl-2016- 000114-egy).

**Table 3.** List of heavy rainfall events in the North Sinai Peninsula at different locations.


Among all the events that occurred in the period 2000–2015, the following episodes that affected the whole Peninsula, although with different intensity, have been selected and analyzed: 8 January 2013 and 9 March 2014. The 2013 event started on January 6th with light rainfall and reached a

peak on January 7th. On January 8th, the storm continued only along the north coast of North Sinai, covering mainly the north Sinai and parts of south Sinai, and the maximum daily rainfall (29.8 mm) was observed at Rafah station. The 2014 event lasted over Egypt for three days from March 7th until March 9th. This event affected all Sinai with a maximum daily rainfall of 30 mm measured at Dahab station. Table 4 summarizes the total amount of precipitation that occurred in Upper Egypt, the Red Sea, and Sinai Peninsula during the storm (7–9 March 2014), while on the 10th it moved rapidly towards East. The evolution of the 2014 storm as captured by TRMM satellite images is shown in Figure 8, where it is clear how the storm, which started to form in Upper Egypt on the 7th, then rapidly moved toward the North-East and strongly affected Sinai Peninsula in the following days (8th and 9th of March).

**Figure 8.** Episode 7–10 March 2014. Storm evolution over Sinai and surrounding region. (Source: TRMM satellite data).

Results of the analysis of the large-scale configuration during these episodes show some similarities, with heavy rainfall usually resulting from mesoscale convective systems embedded in synoptic-scale disturbances interacting with the complex terrain of the Peninsula. The evolution of the storm (not shown) is similar in the selected cases (as an example, Figure 9 shows the spatial distribution of the geopotential at 700 hPa during the selected episodes with a low pressure system over the Eastern Mediterranean bringing a disturbance to the region of interest, which then moves eastward), and a correlation is also noticed with the atmospheric circulation over the Arabic Peninsula and with low-pressure systems over the Eastern Mediterranean. In addition, the distribution of the rainfall between the Northern and Southern parts of the Peninsula is strongly affected by the complexity of the terrain and by the position of the mountain ranges, which is strongly influencing the areas where the majority of rain is observed.


**Table 4.** Total rainfall in Upper Egypt, the Red Sea, and the Sinai Peninsula during the 7–9 March 2014 event.

**Figure 9.** Geopotential height at 700 hPa for the selected episodes: 8 January 2013 and 9 March 2014. (Source: NCEP-NCAR reanalysis).

The mechanism leading to storms with torrential rainfall, thunder, and lightning has been studied by some authors [21] and references therein [18,27,28]; in these works, it is emphasized that they are phenomena of very limited spatial and time extent. Although this part of the study is not yet fully completed, some general remarks are outlined here. Climatically the whole Levant is under the influence of the westerly winds, i.e., on average, the winds blow from the west towards the east and their intensity increases with altitude, and the weather is usually characterized by the presence of a Red Sea Through (RST), i.e., a low-level pressure extending from the African Monsoon over Equatorial Africa, northward over the Red Sea region toward the Eastern Mediterranean (EM). The RST, generated and supported by thermal forcing and supported by the presence of the local complex topography, strongly influences the weather in the whole Levant, usually bringing dry and hot conditions.

The size of storms moving in the eastern part of the Mediterranean basin is of the order of a few km, and their lifetime is only few days. Many storms in the East Mediterranean originate in the region around Cyprus; these storms travel eastwards. Therefore, they do not usually hit Egypt. However, if the conditions are favorable in the Red Sea, they are diverted southeastwards, reaching North Egypt; if their lifetime is sufficiently long, they can hit Central Egypt. These storms are usually associated with the presence of an Active Red Sea Trough. While a (non-active) Red Sea Trough (RST) is a low-level trough extending northward from East Africa over the Red Sea region towards the Levant that brings dry, hot weather, an Active Red Sea Trough (ARST) is an RST accompanied by an upper trough which brings severe weather, heavy precipitation, and possible flash floods as in the case studies analyzed.

In a very simplified scheme, several factors can contribute to transform an RST into the active Red Sea storm; the presence of these factors enhances the tropospheric instability and the synoptic-scale forcing and inducing ascending motions with the formation of an ARST. Usually, the ARST configurations are more active in autumn, when there are two coinciding favorable elements: the position of the African Monsoon and Subtropical Jet Stream (STJ). These elements are absent in winter and spring, when the Middle East (ME) can be affected by extreme rainfall resulting from tropical-extratropical interactions. In the beginning, the systems, which started over Egypt, presented characteristics very similar to a Mesoscale Convective System (MCS), with rainfall particularly intense during its first stage when the convection was dominant. After this first phase, the intensity of rain, as well the role of convection, diminished considerably while the storms move eastwards. The kind storms weather affects not only Egypt but also the Middle East.

#### *3.3. Expected Future Climate Changes*

Considering the fact that regional climate information is needed for decision-making on societal issues such as vulnerability and adaptation to a changing climate with weather/water extremes, the Coordinated Regional Climate Downscaling Experiment (CORDEX: https://cordex.org/) has been developed. CORDEX is a World Climate Research Project (WCRP) framework to evaluate regional climate model performance through a set of experiments aimed to the production of regional climate projections [29]. The CORDEX vision is to advance and coordinate the science and application of regional climate downscaling through global partnerships, and its main goals can be summarized as follows:


Thanks to CORDEX, a set of experiments are available for different domains, including the Middle East-North Africa (MENA) domain, which contains the region of interest of the present study [30,31], and few selected results are briefly discussed here. Figures 10 and 11 show, respectively, the expected changes in temperature and precipitation over a domain including Egypt and the Arabian Peninsula for the control period 1971–2000 and for three future time periods: 2011–2040, 2041–2070, and 2071–2100 for an intermediate stabilization Representative Concentration Pathway (RCP4.5). These figures show the result of an ensemble of nine different Global Circulation Models (GCMs), in which a general increase in the mean temperature for the mid-to-end of the century is evident. This will occur all over the country except for a narrow region along the Mediterranean Sea, and there will be a decrease in the mean annual precipitation for the same periods; this pattern is more evident in the Eastern Desert and Sinai Peninsula.

Focusing on the Sinai Peninsula, several studies [32–34] showed that several parameters affected the occurrence of the flash floods over the Sinai Peninsula, namely urban expansion and extreme rainfall events. Three ensembles from three Global Circulation Models (GCMs)—CNRM-CM5, EC-EARTH and MPI-ESM-LR—downscaled by the Regional Climate Model (RCM) were used to present the future change in temperature and rainfall over Sinai through the mid (2041–2070) and the end of the century (2071–2100), where both periods are compared with the baseline period (1971–2000). All three ensembles are obtained from CORDEX data, with two ensembles (CNRM-CM5, EC-EARTH) from MENA Domain (created by RICCAR Project: https://riccar.org/) and one ensemble (MPI-ESM-LR) from Africa Domain. The ensembles analyzed are for one emission scenario (RCP4.5) with resolution 50 km × 50 km. Figures 12 and 13 show the future change in temperature over Sinai during the periods of 2050s and 2080s, respectively, while Figures 14 and 15 show the future change in precipitation over Sinai during the same periods. The results show increase in temperature for all three ensembles, while the change of temperature will increase highly during the end of the century (2071–2100). In addition, winter will become warmer, especially in mid and south Sinai, with increases in temperature between (0.2 ◦C to 3 ◦C) for both future periods, 2050s and 2080s. It is not clear from the results which model

give the worst result, but all of them agree with the increase in temperature. The precipitation will be affected: Where the results show that the rainfall will be shifted to the summer, sudden storms will be expected to occur in summer during the mid and end periods (2041–2100). The results show increases in amount of the precipitation may reach to 300% in some local areas. There is no change expected during winter and autumn; however, south Sinai will suffer by decreasing rainfall during autumn. The expected distribution of the rainfall shows the localization, which indicates increases in damage (see Figures 14 and 15).

**Figure 10.** Temperature. Results from an ensemble of nine GCM, for RCP4.5 scenario. Ensemble mean (mm/day) for the control period (1971–2000) (upper left panel-color scale on the upper left). Change in Ensemble mean (%-color scale at the bottom), for 2011–2040 (upper right), 2041–2070 (bottom left) and 2071–2100 (bottom right) compared with 1971–2000. (Source: https://cordex.org/).

**Figure 11.** Precipitation. Results from an ensemble of nine Global Circulation Models (GCMs), for RCP4.5 scenario. Ensemble mean (mm/day) for the control period (1971–2000) (upper left panel-color scale on the upper left). Change in Ensemble mean (%-color scale at the bottom), for 2011–2040 (upper right), 2041–2070 (bottom left) and 2071–2100 (bottom right) compared with 1971–2000. (Source: https://cordex.org/).

**Figure 12.** Temperature. Results from an ensemble of three GCMs for RCP4.5 scenario. Change in Ensemble mean monthly (C) for the future period 2041–2070 (2050s) compared with 1971–2000. January, April, July, and October are examples presenting winter, spring, summer, and autumn, respectively.

**Figure 13.** Temperature. Results from an ensemble of three GCMs for RCP4.5 scenario. Change in Ensemble mean monthly (C) for the future period 2071–2100 (2080s) compared with 1971–2000. January, April, July, and October are examples presenting winter, spring, summer, and autumn, respectively.

**Figure 14.** Precipitation. Results from an ensemble of three GCMs for RCP4.5 scenario. Change in ensemble mean monthly (no sign where the change is produced by dividing future by baseline) for the future period 2041–2070 (2050s) compared with baseline period 1971–2000. January, April, July, and October are examples present winter, spring, summer, and autumn, respectively. (White color indicates that baseline value equals zero).

**Figure 15.** Precipitation. Results from an ensemble of three GCMs for RCP4.5 scenario. Change in ensemble mean monthly (no sign where the change is produced by dividing future by baseline) for the future period 2071–2100 (2080s) compared with baseline period 1971–2000. January, April, July, and October are examples present winter, spring, summer, and autumn, respectively. (White color indicates that baseline value equals zero).

#### **4. Conclusions**

The Sinai Peninsula is a region characterized by severe data limitations and by the occurrence of thunderstorms very limited in time and space; therefore, the improvement of Weather Prediction Systems feeding Early Warning Systems is difficult. The lack of reliable WPS and appropriate EWS impair not only the design and adoption of adaptation and mitigation measures but also the development of emergency plans to save lives as well as the harvesting of water precious for the local Bedouin Communities, which is otherwise lost.

The scientific problem addressed in the paper is to assemble information about severe thunderstorms leading to flash flood in the two sub-regions, North and South Sinai, and to present a first attempt to describe, for the same region, the expected evolution of climate in the future. The study presented here is a step forward in order to respond to the need of a deeper knowledge of extreme rainfall in Sinai Peninsula that is driving flash floods affecting population and infrastructure with damage and elevated costs.

The results presented are derived from the Academy of Scientific Research and Technology of Egypt (ASRT) and the National Research Council of Italy (CNR) joint project and show, in particular, the intensity of the current climate change in terms of increase in temperature, decrease in precipitation, and increase (in number and intensity) of heavy rainfall episodes in the interested region. Some results were also discussed showing an increase in the severity of the changes in temperature and precipitation under future climate scenario for an intermediate stabilization Representative Concentration Pathway, RCP4.5.

Further analysis of the heavy rainfall episodes generating flash floods in both the North and South part of the Peninsula is not only needed but also particularly useful to increase the knowledge about the generation and evolution of these short-lived and patchy storms, focusing, in particular, on the atmospheric factors driving their formation. A deeper knowledge of the mechanism could also provide some indication for improving the weather forecast systems over the region, while results will contribute to the development of regional climate scenarios at higher resolution to be used for the implementation of more sophisticated risk maps for future decades. Results can also encourage and sustain the elaboration of adaptation plans for population, agriculture, and water resources management and help decision-makers to plan the construction of water harvesting structures [35].

In conclusion, the knowledge of extreme rain events in Sinai is presently scarce; thus, this study intends to give a contribution to the improvement of their understanding. A possible future study to address this could be a comparative analysis of episodes occurring in the Sinai Peninsula and in neighboring countries, because thunderstorms affecting the Peninsula then move eastward and largely impact the whole Levant.

Further specific studies on the Sinai's present and future climate that can help to understand better if there will be an evolution of heavy rainfall episodes driving flash floods are needed. Sophisticated risk maps for future decades are also needed, together with adoption of adaptation plans for population, agriculture, and water resources management. Finally, results from this and future studies will be largely used by decision-makers in their design and planning of the construction of water harvesting structures in the near future for the good of the Sinai people.

**Author Contributions:** Individual contributions to the research is as follows: Conceptualization, M.B. and D.A.; methodology, M.B. and D.A.; formal analysis and investigation, M.B., D.A., I.S.A.Z., and G.D.; writing—original draft preparation and review and editing, M.B.; writing—review and editing, D.A., I.S.A.Z., and G.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by ARST and CNR in the framework of the Agreement on Scientific Cooperation between the Academy of Scientific Research and Technology of Egypt (ASRT) and the National Research Council of Italy (CNR), who provided support for scientists' mobility.

**Acknowledgments:** The study has been funded in the framework of the Agreement on Scientific Cooperation between the Academy of Scientific Research and Technology of Egypt (ASRT) and the National Research Council of Italy (CNR). The two Institutes participating to the project were, respectively: the Water Resources Research Institute (WRRI) of the National Water Research Center (NWRC), and the Institute of Biometeorology (CNR-Ibimet, now CNR-IBE). M.B and G.D. are grateful to the research group at the WRRI-NWRC for their kind hospitality and useful discussions and to the Italian Scientific Attaché Office in Cairo for support. D.A. and I.S.A. are also thankful to the research staff of the CNR-IBE for supporting the research stay in Italy that enabled the completion of this work.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Floodplain Settlement Dynamics in the Maouri Dallol at Guéchémé, Niger: A Multidisciplinary Approach**

#### **Andrea Galligari 1,\*, Fabio Giulio Tonolo <sup>2</sup> and Giovanni Massazza <sup>1</sup>**


Received: 12 June 2020; Accepted: 9 July 2020; Published: 13 July 2020

**Abstract:** In Sahelian Africa, rural centers have been hit by catastrophic floods for many years. In order to prevent the impact of flooding, the flood-prone areas and the settlement dynamics within them must be identified. The aim of this study is to ascertain the floodplain settlement dynamics in the Maouri valley (135 km2) in the municipality of Guéchémé, Niger. Through hydraulic modeling, the analysis identified the flood-prone areas according to three return periods. The dynamics of the settlements in these areas between 2009 and 2019 were identified through the photointerpretation of high-resolution satellite images and compared with those in the adjacent non-flood-prone areas. Spatial planning was applied to extract the main dynamics. The synergic application of these disciplines in a rural context represents a novelty in the research field. Since 2009, the results have shown a 52% increase of the built-up area and a 12% increase in the number of buildings, though the increase was higher in the flood-prone areas. The factors that transform floods into catastrophes were identified through perceptions gathered from the local communities. Three dynamics of the expansion and consolidation of buildings were observed. Specific flood risk prevention and preparation actions are proposed for each type of dynamic.

**Keywords:** building consolidation; extreme precipitations; flood exposure; flood risk; satellite remote sensing; settlement dynamics; sustainable rural development; vulnerability

#### **1. Introduction**

In recent years, floods in Sahelian Africa have become more frequent and catastrophic [1–3]. Over the last decade, flood-related issues have begun in this area as a result of the unusual magnitude of climatic events [4–6]. Extreme rainfall and its effect on an increasingly altered land surface area are the main flood-driving factors usually reported in the literature [7,8]. Arboreal and shrub cover has been greatly reduced to satisfy the incessant need for arable land [9] and firewood for a continuously growing population. The change of land cover has exposed the soils to degradation processes [10] that increase the runoff [11,12] and lower the rainfall threshold, which, in turn, causes related damage.

Theimpact of hydrometeorological eventsis aggravated byinsufficient rainwatermanagement [13–17] and water-resistant homes. Anthropic pressure in rural areas [18] has led many people to settle within the flood-prone area (FPA) in the absence of regulation and monitoring by the authorities [19]. Flood damage further exacerbates poverty and risks nullifying investments in the primary sector. This sector in the Sahel is closely linked to the reduction of poverty, food insecurity, and inequalities [20].

In rural Sahel, prevention and preparation for flooding through local plans is struggling to take hold due to a lack of financial resources [21], human resources [22], and systematic and detailed knowledge on the exposed settlements [23]. The recourse to low resolution remote sensing and digital elevation models (DEM), photointerpretation with semi-automatic classification algorithms [24], or with urban growth simulation models [25] does not achieve the level of detail on the assets and exposed population that is needed on a local scale. The use of hydraulic numerical models is still occasional [26] and reveals different application fields and different precision levels [27–29]. FPA precision depends on the availability and quality of topographic and hydrological observations [30]. The application field ranges from flood assessment to early warning systems [31], hydraulic constructions [32], and eco-hydraulics [33]. As a result, the literature still offers little on floodplain settlement dynamics and the expansion of settlements is often not quantified [34,35]. This is the first gap that must be addressed in order to improve flood disaster prevention. The second gap concerns the durability of houses in the event of flooding. Currently, the focus is mainly on the precariousness of urban construction [36–39], but less on consolidation [40] in the rural context, which increases the capital exposed to flooding. Once again, here, it is essential to know the characteristics of the housing stock in order to identify the retrofitting measures that can make buildings more resistant to water [41].

The objective of this study is to ascertain the changes of settlements and housing stock in the FPA over the past decade by answering two questions: first, does demographic growth in a rural context increase the exposure of settlements to flooding? Second, does the improvement of building materials increase the flood resistance of homes? Knowledge of these dynamics could help local administrations to identify and localize flood prevention and preparation measures [42].

The synergic contribution of three different disciplines (remote sensing, hydraulic, and spatial planning) and their application in a rural context represents a novelty of the research. To simplify the interpretation, the acronyms adopted in this study are listed in Table 1.


**Table 1.** Glossary of acronyms used in the text.

#### **2. Materials and Methods**

#### *2.1. Study Area*

For this study, a rural area exposed to floods in one of the countries with the highest rural population growth in the Sahel was considered: the municipality of Guéchémé (109,000 inhabitants in 2012) in Niger. Between 2001 and 2012, the municipality's population grew by 23%. The municipal territory is in a watered region (annual average rainfall of 560 mm) of a semi-arid country [43,44]. The studied area (135 km2) is the Maouri Dallol (meaning "valley" in the Fula language), a floodplain that is densely cultivated thanks to the presence of water, but partially exposed to a flood hazard. The floodplain accommodates 87 settlements consisting of 23,000 inhabitants, with one of the country's highest densities (167 inhabitants/km2) [45] (Figure 1).

**Figure 1.** Territorial framework of the Maouri Dallol in the municipality of Guéchémé: Dallol river channel (1), study area (2), main settlements (3), municipal border (4), national border (5), national road no. 1 (6), unpaved tracks (7), and weather stations (8).

Primarily, the study determined the past trend of rainfall and its impact and identified the FPA according to three return period scenarios. Subsequently, it analyzed the settlement dynamics between 2009 and 2019 by identifying the built-up areas and type of buildings based on high-resolution satellite images (Figure 2).

**Figure 2.** Methodology flowchart.

#### *2.2. Rainfall Data Analysis*

The analysis of the settlement dynamics was based upon the synergic contribution of different disciplines. The first phase used elements of climatology for the pluviometric characterization of the territory. A series of 38 years of daily rainfall (1981–2018) registered at two weather stations of the National Meteorology Directorate (DMN, according to the French acronym) was used to track the local rainfall profile (annual accumulation) and the intense rainfall within the 95th percentile (trend) [46]. That trend was compared with the database of flood damage recorded and made accessible by the Niger flood database (BDINA, according to the French acronym) [47]. In addition, in 2018–2019, two meetings were organized with the municipality and with the community of Guéchémé to ascertain the local perception of heavy rainfalls, their consequences, and the factors that transformed them into a catastrophic event.

#### *2.3. Hydrological Analysis and Hydraulic Modeling*

The second phase applied hydraulic methods to identify the FPA. The upstream hydrographic basin of the study area was defined through the DEM, the hydrological data and maps based on shuttle elevation derivates at multiple scales (HydroSHEDS) hydrographic grid [48,49], and GIS software. The morphology was derived from a 90 m DEM, which was resampled and merged with a 10 m DEM for the North part of the study area [50]. The concentration time was computed through Giandotti's formula according to the size and characteristics of the catchment area and the river channel [51–53]. Using the rainfall data, the maximum rainfall, meaning the maximum precipitation during the concentration time—i.e., the period that generated the maximum runoff in the outlet point of the considered basin—was calculated.

The statistical analysis of the annual maximum rainfall was conducted through the probability density functions (PDF) commonly applied to hydrology, such as log-normal, exponential, Gumbel, and generalized extreme value distributions [54,55]. Anderson–Daring and Pearson tests were used to identify the distributional adequacy of PDF and to ascertain the best fit with the dataset [56]. The statistical analysis facilitated the determination of the critical precipitation according to the annual probability of occurrence (*P*) and the computation of the relative return periods (*RP*), which were equal to the inverse of the annual probability (*P*):

$$R\_P = 1/P \tag{1}$$

The *RP* values chosen for this case study were 2, 20, and 200 years, corresponding to events with high, medium, and low probability of occurrence.

These return periods were adopted to ensure the maximum variability and a clear distinction of the three scenarios, which were quite close due to the flat morphology of the large floodplain and the DEM inaccuracy [57].

Hydrographs and maximum discharges were computed through a hydrological model covering the upstream basin. The model was developed on the Hydrologic Engineering Center—hydrologic modeling system (HEC-HMS) software according to the curve number (CN) method, the meteorological data recorded in the Guéchémé gauge station, and the endorheic configuration of the catchment area [58–60]. The CN was defined according to the land cover identified from the Copernicus dataset [61–63]: the watershed was almost completely covered by herbaceous vegetation (54%), sparse vegetation (24%), cropland (13%), and shrubs (8%). No discharge data were available for calibration. A validation was conducted based on the literature values of the runoff coefficient for the Sahelian area [64].

The determination of the FPA was carried out through a monodimensional hydraulic numerical model realized along 50 km of the Maouri Dallol floodplain [65]. The model was created with the Hydrologic Engineering Center—river analysis system (HEC-RAS) software [66] and was opportunely extended upstream and downstream to avoid anomalies related to boundary conditions. The hydraulic model was based on the composed 90 m and 10 m DEM morphology, the typical roughness of the vegetation present in the rainy season, and the discharge resulting from the hydrological analysis [67]. No surface water level observations were available for calibration. A validation of the flooded areas was conducted with the in-situ observations taken during the survey in July, 2018 and the ground-water levels of the piezometers of the study area.

#### *2.4. Satellite-Based Settlement Dynamics*

The third phase consisted of identifying the dynamics of the built-up area and the type of buildings (durable, non-durable) in the FPA and in the surrounding areas in 2009 and 2019 through high resolution satellite image analysis. This phase was developed in three main steps.

First, the opportunity of extracting single buildings through semi-automatic elaborations of high-resolution satellite images was verified, testing different typologies of classification algorithms. Images were acquired between 6 and 14 October, 2009 [68] and between 1 and 6 September, 2019 [69], as detailed in Table 2.


**Table 2.** Satellite imagery specifications.

<sup>1</sup> Multispectral and panchromatic dataset.

The images were subsequently pre-processed radiometrically (pansharpening [70] and radiometric calibration [71]) and geometrically (mosaicking and orthorectification [72]). The classification tests with pixel-based algorithms [73] (spectral angle mapper (SAM)), object-oriented algorithms [74] (image segmentation), and spectral indices (normalized difference built-up index (NDBI)/normalized difference vegetation index (NDVI)) [75] highlighted the difficulty in distinguishing, in a semi-automatic way, the roofs from the roads due to the very similar material used (respectively, earth and laterite) with overlapping radiometric responses.

Consequently, the second step consisted of photointerpretation by an image analyst. The inhabited areas of the riverbed were determined, and the buildings were subsequently digitalized in 2009 and 2019 through computer aided photo interpretation (CAPI). The use of a GIS (geographic information system) environment made it possible to generate a database containing not only the building geometries, but also the ancillary information, such as the material used for the roofs (straw, mud, and corrugated iron sheets, which constitute as many stages of the consolidation of constructions). The delimitation of the built-up areas took into account contiguous building lots (identifiable with walls or hedges) and non-contiguous building lots less than 50 m away from the contiguous area. Twelve settlements with significant building expansion in the FPA, and as many in the non-flood-prone areas (NFPA), making a

total of 24, were selected as case studies. The settlements toponyms were extracted from the national repertoire of inhabited locations.

The third step superimposed the built-up area with the FPA according to three return period scenarios, allowing for an automatic multi-temporal analysis of the changes occurring between 2009 and 2019 in terms of the share of built-up area compared to the total area of each settlement and the share of buildings with roofs made from corrugated iron sheets compared to the total buildings of each settlement.

The lack of precise data for the calibration of the model was the main limitation of this FPA characterization: it ensured a precision that was higher than GIS flood hazard mapping [76] but lower than state-of-the-art hydraulic models [77]. The low accuracy achieved with standard semi-automatic building extraction methods from satellite images was also a limitation, but was bypassed by means of manual digitization.

#### **3. Results**

In the past, the Maouri Dallol was crossed by a stream that originated in Northern Niger and flowed into the Niger River 910 km downstream. Today, its regime is intermittent and its endorheic behavior feeds a wetland with an aquifer close to the surface [78]. Heavy rains flood the valley both due to surface flow and water table rise. In Guéchémé, the Dallol crosses one of the most densely populated zones of Niger. Of the 87 settlements identified in 2019, one is large (Angoual Chekaraou), six are medium, 21 are small, and 60 consist of a few houses (Table 3).


**Table 3.** Settlements in the Maouri Dallol of Guéchémé by size in 2019.

#### *3.1. Climatic and Hydraulic Characterization*

In the 1980s, Guéchémé had a dry period, followed by a wet decade and then an alternation of wet (with a declining trend) and dry years. The accumulated rainfall remained above 400 mm per year, a critical threshold for dry farming [79–81], except in the years 1984, 2016, and 2017 (Figure 3). One of the most significant changes in rainfall over the last forty years has been the increase in value (expressed in mm) of rainfall falling in the 95th percentile, or extreme rainfall. These precipitations increase from 40–50 mm/day (in the nineties) to today's 60–70 mm/day (Figure 4).

The hydrologic analysis revealed the following results: (1) the concentration time of the Dallol Maouri watershed was 210 h (about 9 days), (2) the only PDF that passed the adequacy distribution tests was the log-normal one, and (3) the cumulated rainfall, based on the concentration time of 9 days for the computed return period (RP) ranged from 133 to 287 mm. The flood hydrographs, computed with the hydrological model, agreed with the hydrology of this endorheic area.

The FPA extended between 25 and 67 km2, depending on the flood scenario considered (Table 3). The FPA covered a considerable surface area due to the high transverse extent of the floodplain at Guéchémé (approximately 4 km), characterized by a considerable number of counter slopes. The NFPA surrounding the RP200 flood limit, considered as a reference to compare the settlement dynamics, extended over 68 km2 (Figures 5 and 6).

The hydraulic model also enabled the determination of flow velocity and water depth, which could be valuable results for evaluating the effects on the assets concerned. The flow velocity was quite low due to the low longitudinal slope (0.04% = 40 cm/km) of the riverbed, with a maximum value of 0.85 m/s in the RP200 scenario. By contrast, the water depth was quite considerable, reaching a maximum value of 1.78 m in the RP200 scenario (Table 4).

**Figure 3.** Annual accumulated rainfall (blue) (Guéchémé and Guéchémé Centre de Santé stations) between 1981 and 2018, critical rainfall threshold of 400 mm/year (red), incomplete data (grey).

**Figure 4.** Trend of precipitations in the 95th percentile (red) (Guéchémé and Guéchémé Centre de Santé stations) between 1981 and 2018 and incomplete data (grey).

**Table 4.** Extension and characteristics of the flood-prone areas in the Maouri Dallol based upon the return periods (RP): velocity (V) and depth (D) referred to the urban area distribution.


**Figure 5.** Flood scenarios in the Guéchémé Dallol with probability of occurrence: high RP2 (1), medium RP20 (2), low RP200 (3), study area (4), main settlements (5), Maouri Dallol river channel (6), unpaved tracks (7), hydrographic basin limit (8), and creek flow direction (9).

**Figure 6.** Close-up view of the flood scenarios in the Guéchémé Dallol with probability of occurrence: high RP2 (1), medium RP20 (2), low RP200 (3), study area (4), main settlements (5), unpaved tracks (6), and Maouri Dallol river channel (7).

#### *3.2. Local Perception on Floods*

The meetings with the local authorities and the community of Guéchémé identified heavy rainfall, the reduction of vegetation, and the degradation of the soil as drivers of the increased runoff. These, together with the water table rise, caused floods in the Maouri floodplain. These floods—particularly the memorable ones of 1994, 2012, 2015, and 2016—led to the collapse of many buildings. These events are substantially reflected in those listed in the BDINA database relating to the period 2007–2016. In 2012, 2015, and 2016, many collapsed buildings were reported, particularly in 2013 (Table 5). These events also caused extensive damages to cultures, but this aspect was not considered in this research.


**Table 5.** Catastrophic flood comparison between community remembrance and the Niger flood (BDINA) database between 2012 and 2016.

#### *3.3. Settlement Dynamics*

In ten years, the number of settlements in the study area has increased by 7%: 13% in the NFPA and stable in the FPA; in 2009, 45 settlements were in the NFPA and 36 partially or entirely in the RP200 FPA; ten years later, 51 settlements were in the NFPA and 36 settlements in the RP200 FPA (Table 6, Figure 7).

**Table 6.** Evolution of the number of settlements in the study area between 2009 and 2019.


**Figure 7.** Settlement localization in the Maouri Dallol of Guéchémé: in non-flood-prone areas (1), in flood-prone areas (2), flood scenario RP200 (3), study area (4), municipal border (5), national border (6), unpaved tracks (7), and indication of analyzed settlements.

Of the 87 settlements present in 2019 in the study area, 24—split equally between the FPA and the NFPA—recorded an expansion in surface area greater than 25%. This sample, analyzed in-depth, revealed a 52% increase, particularly in the FPA (+71%) compared to the NFPA (+30%) of the built-up area. The number of buildings increased by 12% (0% in the NFPA and 21% in the FPA) (Table 7). The wide differences of increment between built-up area and buildings led to a decrease in building density (from 40.6 to 29.4 build/ha) as a result of a consolidation process through the installation of buildings with corrugated iron sheet roofs. Between 2009 and 2019, these buildings increased by 3% to 21%, without substantial differences between the FPA and the NFPA. The roofs made from corrugated iron sheets thus increased by 592% against a 10% increase in buildings. In some locations, corrugated iron sheets are now used in almost half of the buildings: an increase of 1400% in the decade considered (Table 8).

**Table 7.** Building expansion of 24 settlements (12 in the flood-prone area (FPA) and 12 in the non-flood-prone area (NFPA)) in the Guéchémé Dallol between 2009 and 2019.


**Table 8.** Roof dynamics of 24 settlements (12 in FPA and 12 in NFPA) in the Guéchémé Dallol between 2009 and 2019.


The exposure of buildings to flooding saw an increase of 29% (40% more than the total increase) in the area with a low probability of flooding RP200. At the same time, the exposure of buildings with corrugated iron sheet roofs increased by 695% in the same RP200 area, which was slightly more than the total increase (+609%). The consolidation of buildings was generally higher in the FPA than in the NFPA: the maximum consolidation occurred in the area with a high probability of flooding RP2 (+27%), followed by that with a low probability of flooding RP200 (+12%), and only lastly in the area with a medium probability of flooding RP20 (+9%) (Table 9).

**Table 9.** Flood exposure of buildings in the Guéchémé Dallol between 2009 and 2019, comparison between building expansion and building consolidation in the FPA.


<sup>1</sup> RPx Δ2009–2019/Total Δ2009–2019; <sup>2</sup> RP200 scenario referenced data.

#### *3.4. Main Dynamics*

In the Guéchémé Dallol, in the decade considered, three main settlement dynamics emerged (Table 10, Figures 8–10):


**Table 10.** Settlement dynamics of representative settlements for each main dynamic between 2009 and 2019.


<sup>1</sup> Percentage points.

**Figure 8.** *Cont*.


**Figure 8.** Schematization of (**a**) less dense building expansion with dense pre-existing built-up area; (**b**) building consolidation with reduction in density of the pre-existing fabric due to building replacement; (**c**) expansion and building consolidation with reduction of the pre-existing fabric due to building replacement and new less dense expansions; precarious building (1), semi-permanent building (2), list of representative settlements for each main dynamic.

**Figure 9.** Building expansion in the FPA in Angoual Chekaraou between 2009 (**1a**) and 2019 (**1b**): high (1), medium (2), and low (3) probability of flooding, built-up area (4), buildings with iron sheet roofs (5), unpaved tracks (6), and 2009 built-up area (7).

**Figure 10.** Details of the three main settlement dynamics of straw (1), mud (2), and corrugated iron sheet (3) roofs in Angoual Chekaraou (**a1**,**a2**), Lokoko (**b1**,**b2**), and Toullou (**c1**,**c2**) between 2009 and 2019.

A summary of the settlements data is reported in Table 11 (further information is available in the Supplementary Material).


**Table 11.** Surface areas, buildings, and roofs in FPA RP200 and in the NFPA.

<sup>1</sup> Disappeared between 2009 and 2019. <sup>2</sup> Settlements marked with a '#' are not reported in the National Repertoire of Localities (RENALOC).

#### **4. Discussion**

In three-quarters of the countries South of the Sahara, the rural population today is still in the majority and strongly increasing [9,18,82]. In these countries, the primary sector remains strategic for development and for reducing poverty [20]. This suggests the refocusing of attention from the urban sector [15,26,36] to the rural sector. Sustainable rural development involves protecting settlements from increasingly frequent flooding. Ascertaining if and how far rural settlements are occupying the FPA [3,4,7] should be a preliminary step for identifying appropriate protection and prevention measures: an aspect still investigated little by peer-reviewed literature [83].

The objective of this study was to verify the recent changes occurring in the settlements and housing stock in one of these areas: the Maouri Dallol in the municipality of Guéchémé, Niger. The quantitative analysis of the settlement dynamics between 2009 and 2019 made it possible to achieve this aim thanks to the availability of multi-temporal satellite images. The municipality's valley receives, on average, 595 mm of rainfall per year, slightly more than the regional average of the Dosso region [43,44]: a favorable condition for rain crops in a semi-arid zone, such as the Niger Sahel [79]. However, in the last twenty years, extreme rainfall (95th percentile) has increased by about 20 mm of intensity [6]. The effect of these rains on degraded soil increases the runoff and leads to the greater flooding of the Maouri Dallol. Over the last ten years in the Dallol, settlements have increased in number (+7%) and size.

The first question to be answered in this study was if, in a rural context in demographic growth, the exposure of settlements to flooding had increased: an aspect recognized by literature [2,42,64,84] but rarely quantified [34,35]. It has been ascertained that in the last decade, in the Guéchémé Maouri Dallol, the expansion of the 24 observed settlements occurred for 71% in the FPA and 30% in the NPFA. Therefore, the exposure strongly increased due to human activity.

The second question to be answered concerned the flood resistance of buildings constructed in the last decade in terms of building consolidation. In the Maouri Dallol between 2009 and 2019, there was a marked consolidation of buildings expressed by the transition from mud roofs to those

made from corrugated iron sheets (from 3% to 21% in just 10 years), without distinction between the FPA and the NFPA. This process is decreasing the building density of all the analyzed settlements. The consolidation in the FPA sees an average of +720%, with peaks of up to +3500% in the area with a medium probability of flooding (RP20) in Angoual Chekaraou. This consolidation protects homes from the impact of heavy rainfall on the roof, but not from the runoff and the water table rise. Even though the low flood velocities observed do not represent a decisive factor in terms of the potential damage caused by floods, the considerable flood depths (up to nearly 2 m) pose a real threat to buildings: if water enters homes with mud walls, they are guaranteed to collapse. This demonstrates that the collapse of buildings occurs only because they are built from precarious materials and that concrete materials should be used for the full height of buildings to ensure flood protection.

Many similar studies on floodplain settlement dynamics agree on the process of settlement expansion in flood-prone areas, both in African contexts [85] and in other regions of the world [12], but a peculiar characteristic of the Maouri Dallol is a decreased building density of the settlements compared to other contexts [86]. The consolidation of buildings and its implications on flood exposure is still poorly explored in the Sahel. However, in a broader context, the water effects on buildings after a flood event is fairly recognized [39,41].

One of the interpretative hypotheses of the two observed phenomena is that in the context of a growing population in a rural area and the consequent increased need to put more and more land into cultivation, the new nuclei need to settle close to the fields (and water) during the agricultural season, despite the occurrence of flooding [8]. The consolidation of residences with corrugated iron sheet roofs is undoubtedly an improvement to which everyone aspires as soon as their economic conditions allow it. However, in the event of flooding, the invested capital is also at risk of being lost. The meetings with the local authorities and with the community of Guéchémé highlighted the clear local perception of rainfall changes and the degradation of the soil, but not the increase in the extent of the FPA due to the ever more frequent extreme events. These areas, which are today estimated to be low in flooding probability, may become likely to be flooded if the processes of erosion and soil degradation continue in the future. Precisely for this reason, knowing the FPA according to different rainfall intensities and the settlement dynamics in place within them becomes important for deciding upon the measures to be adopted in the municipal development plan, particularly if extreme events grow in intensity and the population exposed to them increases [23]. The seasonality and high variability of surface flows in the Maouri Dallol may be one of the factors that determines a lack of adaptive responses by the local population compared to watercourses with a more regular regime, as observed in the Sirba River watershed in Niger [65].

The urgency to act is also dictated by the speed (just a decade for the discussed case study) and extent to which the expansion and consolidation of inhabited areas in the FPA have occurred. The municipal development plan could include flood risk prevention (information on at-risk areas, resettlement of exposed inhabitants, or protection of inhabited areas). The three main dynamics among the settlements of the Guéchémé Dallol facilitate the identification of specific measures to reduce the flooding risk. Delocalization or retrofitting of the individual buildings with entry barriers and raised basements is recommended for settlements that spread in low density in the FPA, while the construction of embankments and dikes to protect against runoff water is more suitable for settlements that are still dense [87].

Among the results of this study, the role of satellite remote sensing, fundamental for multi-temporal analysis (in the absence of other reliable reference data: e.g., cadastral data), is highlighted, thanks to the now well-established availability of historical archives (starting from around the year 2000) of high-resolution satellite images. On the other hand, low thematic accuracy has been obtained in the application of semi-automatic standard classification algorithms for extracting building footprints from high-resolution satellite images. These accuracies are undoubtedly affected by the geographical context, characterized by the presence of numerous buildings with mud-covered roofs (or temporarily used for forage conservation) and unpaved roads, both having similar radiometric responses and, therefore, difficult to isolate. Conversely, more promising results are observed for corrugated iron sheet roofs, which can easily be isolated. These results led to an approach based on photointerpretation and manual digitization in a GIS environment. In order to increase the level of the automation of this phase of the methodology, the use of image segmentation techniques based on deep learning algorithms is considered extremely promising [88], also by exploiting the better spectral resolution of recent satellite sensors.

The main limit of the analysis is the accuracy of the perimeter of the FPA. This is affected by the lack of detailed information on both soil morphology and hydrology. In particular: (1) the low planimetric resolution and vertical accuracy of the DEM; (2) the lack of surface flow measurements; and (3) the lack of surface flow measurements makes the determination of the FPA inaccurate. For point (1), the use of high-resolution satellite stereoscopic pairs, which facilitate the extraction of DEM with planimetric resolutions and vertical accuracy in the order of a few meters, is considered promising. For points (2) and (3), it is hoped that this initial characterization of the FPA of the Maouri Dallol floodplain will increase the authorities' awareness of the importance of investing time and resources both in field observations and in flood protection measures in this area.

#### **5. Conclusions**

In recent years, the rural Sahel has been struck by catastrophic floods, which have slowed down the already stunted development process. Until now, the literature has explained these catastrophes by studying climate changes and variations as well as alterations of land cover. Knowledge of floodplain settlement dynamics has not advanced in parallel, especially in rural contexts. This study focuses on a 135 km<sup>2</sup> rural area in Niger, with a high population growth and flood risk, during the decade 2009–2019. The settlement dynamics were ascertained through the photointerpretation of multi-temporal high-resolution satellite images and superimposed on the flood boundaries according to three return period scenarios (2, 20, and 200 years).

A significant increase in the intensity of heavy rainfall in the 95th percentile over the last 20 years (+20 mm/day) and four catastrophic floods during the last decade was observed. However, the settlements increased by extension (+52%) and 70% of this expansion occurred in areas with a medium and high probability of flooding. The number of buildings also increased by 21% in the same areas. Many buildings (+592%) were consolidated by installing corrugated iron sheet roofs in place of traditional mud roofs, which could be a proxy indicator of improved economic prosperity. This consolidation process was more intense in areas with a high probability of flooding (up to +800%). However, this improvement is not sufficient to cope with rain-induced floods, as the runoff water can make the walls of the buildings, which are still built from mud, collapse. Therefore, the investment to improve the roof increases the exposed capital but reduces the exposure to damage caused by floods to a limited extent in a context where the hazard has increased.

The significance of these results should draw the attention of the local authorities, who need to identify and implement prevention and preparation measures in the local development plans. The better knowledge and awareness of flood-prone areas and of how settlements develop around them will help the authorities to protect people and assets from natural hazards and facilitate economic development.

Historical satellite images enabled the extraction of detailed building information at the beginning and end of the study period, but the traditional methods of semi-automatic extraction of individual buildings from high resolution satellite images did not produce adequate results in this geographical context. On the other hand, the hydraulic model developed in this study was able to delimit the floodplain settlement dynamics with sufficient accuracy for local planning with respect to the methods currently applied in the Sahel, such as the creation of a buffer around the rivers and the overlay of images of historical floods over low resolution satellite images on the ground. This type of analysis can be replicated in other rural contexts with high population density, allowing local administrations to identify and localize flood prevention and preparation actions.

Finally, the relevance of a multi-disciplinary research group, which allowed for the synergistic integration of spatial planning, hydraulics, and satellite remote sensing skills to extract value-added information in support of decision-making processes, should be underlined.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2071-1050/12/14/5632/s1, Table S1: Twentyfour settlement dynamics in the Maouri dallol at Guéchémé, Niger (2009–2019).

**Author Contributions:** Conceptualization, A.G.; methodology, A.G., G.M., and F.G.T.; software, G.M., F.G.T.; validation, A.G.; writing—original draft preparation, A.G.; writing—review and editing, A.G., G.M., and F.G.T.; All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by DIST-Politecnico and University of Turin, Italy within the ANADIA 2.0 project.

**Acknowledgments:** The authors would like to thank the Italian Agency for Development Cooperation for cofounding the ANADIA 2.0 project that allowed for the development of this study, Katiellou Gaptia Lawan (Directorate National for Meteorology of Niger-DMN), Aliou Tankari (Ministry of Agriculture and Livestock of Niger-MAE), and Maurizio Tiepolo (DIST-Politecnico and University of Turin) for sharing the results of meetings with the Guéchémé municipality with the authors, the Guéchémé community, and for providing new satellite imagery of the area, and Maurizio Tiepolo for his help during the preparation of this article.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
